Psychological measurement is always attended by error. In psychophysics and generally in cognitive psychology, there are no principled theories of the origin of error and it has been assumed that error terms contain little of explanatory value outside of their distributional properties. This is not true. The residuals that are produced during a block of trials in typical sensory and cognitive experiments are correlated in ways that are seen in complex systems in nature. In many cases the residuals form what is known as a 1/f or flicker noise. In physical systems the presence of 1/f noise often reflects a dynamical system that integrates aspects of both order and disorder. In fact, it is theorized to be a general property of adaptive systems. The presence of 1/f noise in human cognition could be an important key to understanding what is involved in forming representations, making decisions, and allocating attention. The first step towards achieving this understanding is to isolate at a functional level the source of fluctuation. This will be done by a careful dissection of the components of psychological measurement into its perceptual demands, task demands, response mappings, and overt response. The second step is to characterize how the dynamic moves forward, whether it is tied to ongoing physiological fluctuations or whether it is activated on a trial by trial basis by the recruitment of whatever resources are used in forming behavioral responses to stimuli. This distinction will be explored by varying the presentation rate of psychophysical probes. Finally, the manner in which attention modulates the formation of residual correlation will be studied by varying the stimulus complexity in visual search and in forced choice discrimination.